Targeting TMS by brain scan beat the ruler-and-scalp method for depression

Choosing where to aim a depression-treating magnet by scanning each patient's own brain wiring worked better than the standard method of measuring across the scalp, in a small randomised trial from Harvard [s1]. Depression scores fell further when the stimulation was steered by individual brain-connectivity data, and the difference was large enough to be worth a bigger trial [s1]. The caveat is size: 40 people, one centre, and a one-month primary readout — enough to justify the next study, not to change practice.

Transcranial magnetic stimulation (TMS) treats depression by delivering magnetic pulses to the left dorsolateral prefrontal cortex, a region on the surface of the brain involved in mood regulation. "Accelerated" TMS (aTMS) compresses many sessions into a few days and has won US clearance with striking early results [s1]. But a basic question has lingered: does it matter exactly where on that region you aim? The conventional answer uses the "Beam F3" method, which locates the spot by measuring across the scalp from fixed landmarks — cheap, but the same coordinates for everyone [s1]. The alternative is to use functional MRI to find, in each individual brain, the precise patch of cortex most tightly wired to a depression-linked circuit.

What the trial did

Researchers at the Center for Brain Circuit Therapeutics at Mass General Brigham and Harvard Medical School enrolled 40 adults aged 22 to 80 with major depressive disorder that was moderate to severe and resistant to previous treatments [s1]. Every participant underwent a 41-minute resting-state functional-connectivity brain scan; then half were randomly assigned to have their stimulation guided by that scan, and half by the scalp-based Beam F3 method [s1]. Participants, the technicians delivering the treatment, and the study physicians were all blinded to which targeting method each person received — an unusually clean design for a device trial [s1].

For the connectivity group, the target was the spot in the left dorsolateral prefrontal cortex most strongly correlated with a published, publicly available "convergent depression circuit," which includes negative connectivity to the subgenual cingulate, a deep mood-related hub [s1]. The primary outcome was the score on the Montgomery-Åsberg Depression Rating Scale (MADRS) one month after treatment, adjusted for each patient's starting score, with further follow-ups every three months for a year [s1].

What it found

Among the 40 participants (55% female; mean age 45.7 years), the median MADRS reduction at one month was 24 points (interquartile range 19 to 28) with connectivity-based targeting versus 18 points (10 to 23) with scalp-based targeting (P=0.02) [s1]. The estimated effect size was 0.8 on a Cohen's d analogue, with a 95% confidence interval of 0.26 to 1.54 — a moderate-to-large advantage, though the wide interval reflects the small sample [s1]. The authors translate this into a "number needed to scan" of 5, meaning roughly one extra responder for every five patients imaged [s1].

The trial also checked whether the personalised targets were meaningful rather than noise. Each individual's target was reproducible when the scan data were split in half (a 4.47 mm distance between the two estimates) and clearly distinct from other people's targets (12.97 mm apart), a difference that was highly significant (P<0.001) [s1]. In other words, the brain spot the method picked was stable within a person and specific to that person.

How to read it

This is a mechanism-confirming trial, not a practice-changing one, and the authors frame it that way: the result "could facilitate cost-effectiveness analyses and help to power a confirmatory efficacy trial" [s1]. Its strengths are real — randomisation, thorough blinding, and a prespecified primary outcome — but 40 patients at a single academic centre is a small base, and the confidence interval around the effect size stretches from modest to very large [s1]. A six-point median difference on a depression scale is clinically meaningful, yet it rests on a handful of participants and needs replication before anyone concludes that the scan earns its cost.

The finding matters most as a signal that the extra step of imaging each brain — expensive and not widely available — may buy real benefit, which is exactly the sort of claim that expensive add-ons in psychiatry usually make without evidence [s1]. Here there is at least a controlled comparison behind it. Related coverage has examined whether TMS works for depression at all, how ketamine compares with electroconvulsive therapy in treatment-resistant depression, and a trial of repeat subcutaneous ketamine for the same condition.

What to watch

The obvious next step is the larger confirmatory trial the authors call for, powered to confirm the effect and to test whether the one-month advantage holds across the full year of follow-up already built into the study [s1]. The economic question runs alongside the clinical one: personalised targeting only makes sense if the added benefit justifies the added scan, and that calculation awaits firmer numbers.

This article describes research and is not medical advice. Decisions about depression treatment are for treating clinicians.

Sources

• Connectivity- vs Scalp-Based Targeting of Accelerated Transcranial Magnetic Stimulation for Depression — JAMA Psychiatry, 24 June 2026